Removing an Inaccessible Node From the Cluster

Removing an Inaccessible Node From the Cluster

If a node has joined the cluster but has communication issues, it might be in a
NotReady
state.
When trying to rejoin, the following failure can occur:
Connection failed. The joining node (IP address of failed remote node) is already known to dqlite (504).
At this point, remove the failed node, then re-add it.

Remove an Inaccessible Node

To remove an inaccessible node, run:
sudo -E /snap/bin/microk8s remove-node --force <NODE NAME>
Replace <NODE NAME> with the node's name; by default, this is its hostname.

Re-add a Node

To re-add the node, follow the instructions in the Secondary Nodes - Setup stepsection of Multi-Node Configuration: Setting up and Installing. Only perform the steps for the node that you are re-adding. Ensure to run the
--finalize
step from the primary node at the end.